R Notebook

library(tidyverse)

Week7 day 1

Dates and Times

Why are they difficult? - formatting them in programming can be tricky -lots of different ways to enter them - tricky data format to work with

ISO-8601 Format - YEAR-MONTH-DAY

Format: YEAR-MONTH-DAY-HH-MM-SS Min digits: 4 b/c of 2000 Clock system: 24 hours Time formats: 4 Timezone - UTC Combined date-times - 2021-12-13-T10:13

library(lubridate)

my_bday <- "09-01-2021"
my_bday

[1] "09-01-2021"

dmy(my_bday)

[1] "2021-01-09"

mdy("January 9th, 2021")

[1] "2021-01-09"

mdy_hms("01-09-21 10:18:26")

[1] "2021-01-09 10:18:26 UTC"

ymd(20190109)

[1] "2019-01-09"

netflix <- read_csv("data/netflix.csv")

netflix <- netflix %>%
  mutate(made_date = make_datetime(year, month, day))

head(netflix)

Add a time

Calculating time spans

• difftime: represents
• period
• durations

Period: Subsetting / breaking down periods into readable / digestable periods

When creating/reporting periods, it is important to be careful/cautious about leap years. Because leap years don’t just appear out of nowhere, it is important to be aware of small changes, i.e seconds/minutes/hours, etc - each year is 365.25 years It is really only important to watch that you get the right amount of days

Netflix promotes new releases for 2 weeks before release Netflix takes shows/movies down after 4 months

netflix <- netflix %>%
  mutate(promo_date = date_added - weeks(2),
         removal_date = date_added + months(4))

netflix %>%
  select(date_added, promo_date, removal_date)

Durations: Records the exact passage of time since a date

netflix <- netflix %>%
  mutate(time_duration = as.duration(today() - date_added))

netflix %>%
  select(date_added, time_difftime, time_years, time_duration)

Timezones

ymd_hms("2021-12-31 11:30:22", tz = "Atlantic/Faroe")

[1] "2021-12-31 11:30:22 WET"

Force_tz - add a timezone, doesn’t change date/time with_tz - figures out what date and time would be if you were in that time zone

Lesson 2

Timeseries

Part of the PDA

• continuous data over time
• time is the primary axis
• append rather than replace

library(tsibble)
library(tsibbledata)
library(feasts)
library(fable)

tsibble is for working with timeseries data feasts - features etc for statistics

index_var - find the index of a tsibble key_vars - find the key of a tsibble

You can use dplyr verbs with tsibble, but you cannot deselect / drop time as a column

vic_year <- vic_elec %>%
   filter_index("2012-01")

Warning: System time zone: "Europe/London".
It may yield an unexpected output. Please see `?filter_index` for details.

Make plots

# calculate a rolling window
# the . in .f corresponds to the current window in the data, i.e. each subset
elec_rolling <- vic_elec %>% 
  mutate(
    temp_moving_avg = slide_dbl(
      .x = Temperature, 
      .f = ~ mean(., na.rm = TRUE),
      .before = 2000,
      .after = 2000,
      .complete = TRUE
    )
  )

elec_rolling

# plot the data 
ggplot(elec_rolling) + 
  geom_line(aes(x = Date, y = Temperature), colour = "grey") + 
  geom_line(aes(x = Date, y = temp_moving_avg), colour = "red")

The slide_dbl() function makes a number of arguments available, so let’s play around with them a little! Try setting .before and .after to 100. How does the plot change? What happens if you set .complete = TRUE?

Feasts package - useful for plotting

key_vars(tourism)

[1] "Region"  "State"   "Purpose"